Transcript Folie 1 - hvonstorch.de

Developing hypotheses about
the variability of climate
variables using Erik den Røde
data – the case of extratropical storminess
Fischer-Bruns, I., H. von Storch, E. Zorita and F. González-Rouco,
2004: A modelling study on the variability of global storm activity
on time scales of decades and centuries. submitted
Empirical evidence about
extratropical storm variability
Estimates
based upon
pressure
readings
Lund and
Stockholm
Bärring and von
Storch, 2004
Estimates based upon repair costs
for dikes in Holland
de Kraker, 1999
Very little
evidence available
ECHO-G
simulations
„Erik den
Røde” (10001990)
and
“Christoph
Columbus”
(1550-1990)
with
estimated
volcanic,
GHG and
solar forcing
Extratropical storminess
• Determined by the
frequency of maximum
wind speeds in a grid cell
of 8 Bft or more (17.2
m/s)
• Number of storm days in
DJF (top) and JJA
(bottom) during
preindustrial period
1550-1850
Extratropical storminess
Pre-industrial: 1550-1850
change from pre-industrial to
industrial period 1850-2000
..
..
a) Mean number of storm days in winter per grid point averaged over the preindustrial and industrially influenced periods of Erik and over the climate
scenario A2 for each hemisphere.
b) Same index as function of time.
c) and d) same, but for North Atlantic region (90W-30E) and North Pacific region
(150E-90W).
Extratropical Storm variations
• North
Atlantic
• Mean nearsurface
temperature
(red/orange)
• storm
frequency
index (blue),
• storm shift
index (green)
• 2 band of
preindustrial
conditions
Storm shift index defined as PCs of storm
frequency EOFs
Extratropical Storm variations
• North Pacific
• Mean nearsurface
temperature
(red/orange)
• storm
frequency
index (blue),
• and storm
shift index
(green)
• 2 band of
preindustrial
conditions
Storm shift index defined as PCs of storm
frequency EOFs
Extratropical Storm variations
• Southern
Hemisphere
• Mean nearsurface
temperature
(red/orange)
• storm
frequency
index (blue),
• and storm
shift index
(green)
• 2 band of
preindustrial
conditions
Storm shift index defined as PCs of storm
frequency EOFs
Conclusions
• During historical times storminess on both
hemispheres is remarkably stationary with little
variability.
• During historical times, storminess and large-scale
temperature variations are mostly decoupled.
• In the climate change scenarios, with a strong
increase of greenhouse concentrations, both
temperature and storminess rise quickly beyond the
2σ-range of pre-industrial variations.
• There are indications for a poleward shift of the
regions with high storm frequency on both
hemispheres with future warming. Altogether, we
have ascertained an increase of the North Atlantic
and SH storm frequency index, whereas the North
Pacific storm frequency index decreases with
beginning industrialization.
Conference on Utility of
multicentury/millenium runs
April, or so, 2006 in …
Madrid
Joint effort of U Madrid, GKSS and MPI, and?
Motivation
• The question of historical reconstructions has
been re-opened.
• All statistical methods (based on regression)
underestimate variability, in particular on longer
time scales as no samples are available for
training regression on these time scales.
• Inflation may help to some extent, but its
evidential basis is based on very few degrees of
freedom.
• Thus, efforts need to combine two sorts of
knowledge, namely empirical (proxy,
instrumental) and conceptual (GCMs).
Issues
• Set-up of multi-century/millium integrations
• Validation of MCMIs
• Utility of MCMIs: Testing diagnostic methods (e.g.,
historical climate reconsructions; non-linear structures)
• Utility of MCMIs: Testing proxy-data inversion methods
by imbedding forward models of proxy formation in the
climate model (e.g., borehole temps)
• Utility of MCMIs: Derivation of hypotheses about
variability of nonärecontructable climate variables (e.g.,
storminess)
We are in a planning stage
- Can change as we like;
- Further partners welcome;
But, don’t worry about organisational work – this will be
(mostly) done by GKSS
Left column:
Leading EOFs of
storm frequency
for the preindustrial period
of experiment H2
for the North
Atlantic, North
Pacific and SH
region (top to
bottom).
Right column:
Corresponding
patterns of linear
slope coefficient
displayed at each
grid point for the
climate change
experiment A2
determined by a
linear trend
analysis.